UEF//eRepository
DSpace https://erepo.uef.fi
Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta
2021
Moisture content variation of ground vegetation fuels in boreal mesic and
sub-xeric mineral soil forests in Finland
Lindberg, Henrik
CSIRO Publishing
Tieteelliset aikakauslehtiartikkelit
© IAWF 2021 All rights reserved
http://dx.doi.org/10.1071/WF20085
https://erepo.uef.fi/handle/123456789/24868
Downloaded from University of Eastern Finland's eRepository
Moisture content variation of ground vegetation fuels in boreal mesic and sub-xeric 1
mineral soil forests in Finland 2
3
Henrik LindbergA,E, Tuomas AakalaB,Cand Ilkka Vanha-MajamaaD 4
5
AHäme University of Applied Sciences, School of Bioeconomy, Visamäentie 35 A, P.O. Box 230, FI- 6
13100 Hämeenlinna, Finland 7
BDepartment of Forest Sciences, Latokartanonkaari 7, P.O. Box 27, FI-00014 University of Helsinki, 8
Finland 9
CCurrent address: University of Eastern Finland, School of Forest Sciences, P. O. Box 111 10
FI- 80101 Joensuu, Finland 11
DNatural Resources Institute Finland (Luke) Latokartanonkaari 9, FI-00790 Helsinki, Finland 12
ECorresponding author: Email: henrik.lindberg@hamk.fi 13
14
15
16
17
18
19
20
21
22
Abstract:Forest fire risk in Finland is estimated by the Finnish Forest Fire Index (FFI), which 23
predicts the fuel moisture content (FMC) of the forest floor. We studied the FMC variation of four 24
typical ground vegetation fuels,Pleurozium schreberi, Hylocomium splendens, Dicranumspp., and 25
Cladoniaspp., and raw humus in mature and recently clear-cut stands. Of these, six were sub-xeric 26
Pinus sylvestris stands, and six mesicPicea abies stands. We analyzed FFI’s ability to predict FMC 27
and compared it with the widely applied Canadian Fire Weather Index (FWI).
28
We found that in addition to stand characteristics ground layer FMC was highly dependent on the 29
species so thatDicranum was the moistest, and Cladonia the driest. In the humus layer, the 30
differences among species were small. Overall, the FWI was a slightly better predictor of FMC than 31
the FFI. While the FFI predicted ground layer FMC generally well, the shape of the relationship 32
varied among the four species. The use of auxiliary variables thus has potential in improving 33
predictions of ignitions and forest fire risk. Knowledge of FMC variation could also benefit planning 34
and timing of prescribed burnings.
35
36
Brief summary: The studied four moss and lichen species were found to dry at different rates, thus 37
having different ignition potential and fire risk. Stand type, and particularly developmental stage also 38
affected the drying rates. The fire risk indices could be improved by using these variables, which 39
could benefit fire prevention.
40
41
Keywords:fire risk, forest fire index, forest type, prescribed burning, Norway spruce, Scots pine, 42
stand structure 43
Running head:Variation in moisture content of ground vegetation fuels 44
45
46
47
Introduction 48
In Finland, forest fires declined during the last century. This decline was particularly steep during the 49
latter half of the century. The average annual burned area in 1950s was about 5,700 ha and in the 50
1970s it had declined to approximately 700 ha (Yearbook of Forest Statistics 1990-1991 (1992). In 51
recent decades, the average annual burned area has varied between 200 and 800 ha, only occasionally 52
exceeding 1,000 ha. The average size of an individual fire is currently about 0.4 ha (Finnish Statistical 53
Yearbook of Forestry 2014). The climatological fire risk in Finland was relatively stable during the 54
last century (Mäkeläet al. 2012), so the decline in fire occurrence is explained by other factors, such 55
as efficiency in fire detection and suppression, and changes in ignition sources, stand structure, forest 56
fragmentation, and vegetation (Päätalo 1998; Wallenius 2011). This is also supported by the 57
difference between the fire regimes of Finland and neighbouring Sweden, where the annual burned 58
area has been higher and large fires frequent (Lindberget al. 2020).
59
Although forest fires do not currently form a major risk to society or property in Finland, they still 60
employ rescue services leading to a need to improve forest fire risk assessment methods. This is 61
partially due to the fact, that although the burned area has been low, the annual number of fires has 62
been about 1,300 in the 21st century (Finnish Statistical Yearbook of Forestry 2014). Thus, the small- 63
sized but frequent forest fires burden regional rescue services and local fire brigades during the forest 64
fire season. Several studies have also predicted that the general forest fire risk in Finland (Kilpeläinen 65
et al. 2010; Lehtonenet al. 2014; Mäkeläet al. 2014) and the risk for large fires (Lehtonenet al.
66
2016) will increase in the 21st century. One way to improve the preparedness of rescue services is to 67
improve the ability to predict potential fire hazard days.
68
The fuel moisture content (FMC) of different fuels is one of the key factors when estimating fire risk.
69
FMC is used to predict flammability, and it is also a factor in models predicting fire intensity and fire 70
spread rate. Most forest fire indices are meteorological and use various weather data to compute 71
indices for assessing fire risk (San-Miguel-Ayanzet al. 2003).
72
Currently, the most widely used fire index system is the Canadian Forest Fire Weather Index System 73
(CFFWIS), which was initially designed for the Canadian boreal forest. Since being published in 1970 74
(Van Wagner 1987), it has gradually been adopted in many parts of the world, including different 75
vegetation zones and fuel types (Dimitrakopouloset al. 2011). The FMC estimation in CFFWIS is 76
divided into three moisture codes: Fine Fuel Moisture Code (FFMC), Duff Moisture Code (DMC) and 77
Drought Code (DC) (Van Wagner 1987). These moisture codes are calculated daily based on air 78
temperature, relative humidity (not in DC), wind speed (only FFMC), and rainfall. Two spread 79
indices are then estimated: initial spread index using wind and FFMC and build-up index combining 80
DMC and DC. The spread indices are then combined to determine the Fire Weather Index (FWI) (Van 81
Wagner 1987).
82
CFFWIS has proven suitable in forests with a flammable duff layer typically consisting of a humus 83
layer and moss cover like, for instance, the black spruce (Picea mariana ) (Mill.) Britton, Sterns &
84
Poggenburg) forests in boreal Northern America (e.g. Zielet al. 2020). Fennoscandian coniferous 85
forests have a similar type of duff structure, and CFFWIS has generally been found to work well there 86
(Granström and Schimmel 1998; Tanskanenet al. 2005).
87
Despite the increasing use of CFFWIS, national fire indices are still commonly used in many 88
countries. In Finland, the forest fire risk is estimated and predicted by the Finnish Forest Fire Index 89
(FFI). FFI was constructed in 1996 to replace the former fire index, which was based merely on 90
statistical correlations between weather variables and the occurrence of fires (Heikinheimoet al.
91
1998). In 1996, Sweden started to use CFFWIS as a national forest fire index system (Sjöströmet al.
92
2019), but the Finnish Meteorological Institute (FMI) decided to develop its own index, partly 93
because CFFWIS was considered unnecessarily complicated with its hierarchical structure, and 94
because it was lacking solar radiation as an explaining variable (Heikinheimoet al. 1998).
95
FFI is based on empirical relationships between weather data and the volumetric moisture content of a 96
6-cm thick layer of forest floor. In short (see Supplement 1 and Vajdaet al. (2014) for details), air 97
temperature values are obtained from the ground weather station network and spatially interpolated to 98
a 10 km ×10 km grid using the kriging method (Venäläinen and Heikinheimo 2003). Evaporation is 99
estimated based on this interpolated data and weather prediction models, and the precipitation is 100
received from weather radars (Venäläinen and Heikinheimo 2003; Vajdaet al. 2014). The index is a 101
continuous variable calibrated to vary from 1.0 to 6.0, 6.0 being the driest. The index has been 102
assigned a threshold value of 4.0, at which point it predicts a volumetric moisture content under 20%.
103
When the index exceeds this threshold, a forest fire warning is announced in public media, which 104
forbids the lighting of open fires. It must be noted that the FFI uses volumetric moisture content 105
values based on non-destructive monitoring of fuels and thus they are not directly comparable with 106
gravimetric moisture content values.
107
In addition to its role in wildfire, FMC plays an important role in prescribed burnings, used in Finland 108
as a silvicultural tool and nowadays also for ecological restoration and management for biodiversity.
109
Because of this, the scope of prescribed burnings in Finland has widened in recent years to a more 110
diverse set of burnings with different ecological aims such as burnings of retention trees, restoration 111
burnings in nature conservation areas and management burnings of sun-exposed and xeric habitats 112
(for details see Lindberget al 2020). The various aims also set diverse targets for fire impact and 113
depth. However, despite the recognized importance of fire for restoration, the overall area of 114
prescribed burns has declined in recent decades (Lindberget al. 2020).
115
116
FMC is one of the most significant factors determining the potential days of prescribed burnings and 117
intended burning depth (Sandberg 1980; Fergusonet al. 2002; Hille and den Ouden 2005; Hille and 118
Stephens 2005). Because of different ecological aims, understanding how FMC develops in various 119
fuels and their effect on fire impact and burning result is necessary. As an example, in silvicultural 120
burnings and burnings on barren habitats, the aim is to decrease the organic layer, which requires a 121
sufficiently low FMC. If the moisture of the ground layer and in some cases raw humus is too high, 122
the burning effects are not fully achieved. In restoration burnings, more various moisture conditions 123
are possible, since more diverse burning results are accepted (Lindberget al. 2020).
124
Boreal ground layer species differ in their structure and growth form which affects their water-holding 125
capacity (Peterson and Mayo 1975; Busby and Whitfield 1978; Pech 1989). The aim of this study was 126
to determine the FMC variation of dominant forest floor mosses and lichens and raw humus in 127
different stands of the two most common forest types in Southern Finland. We analyzed how the 128
moisture content of selected species varied as a function of FFI, and we compared the ability of FFI 129
and FWI to predict the FMC of selected fuel materials.
130
We hypothesize that as clear-cut areas and pine-dominated sub-xeric stands receive more radiation 131
and are more exposed to the drying effect of wind: i) ground vegetation fuels dry faster in clear-cut 132
areas as compared to closed-canopy forests, ii) fuels in pine-dominated forests dry faster than in 133
spruce-dominated forests, iii) varying water holding capacity of studied materials explains the 134
possible differences in their FMC behavior and potential days of ignition.
135
136
Materials and methods 137
Study area 138
139
The study area is located in Southern Finland in the Evo State Forest (Fig. 1) belonging to the 140
southern boreal vegetation zone (Ahtiet al. 1968). The elevation of the study area varies between 141
100-190 meters a.s.l., mean annual temperature in the region is +3.1C, the average annual 142
precipitation is 670 mm, and the growing season 160 days (Juvakkaet al. 1995). The bedrock is 143
mostly orogenic granitoid covered by a thick, stony morainic layer, but glacier sedimented areas such 144
as deltas, sandur deltas and eskers with sand or gravel are also common (Okko 1972). Of the sampled 145
stands, the sub-xeric stands were mostly located in sedimented, sandy soils and mesic stands on sandy 146
or fine sandy moraines (Fig. 1).
147 148
Figure 1 149
150
Experimental design and sampling 151
152
Nearly 90% of Finnish forests are managed commercially (Finnish Statistical Yearbook of Forestry 153
2014). The management is typically done relatively uniformly, including artificial regeneration, 2-4 154
low thinnings, and clear-cutting with less than 3% retention of tree volume (Finnish Forestry, Practice 155
and Management 2011, Kuuluvainenet al. 2019). The stands are thus evenly aged, relatively sparsely 156
stocked and most often dominated by Norway spruce (Picea abies L.) H. Karst and Scots pine (Pinus 157
sylvestris L.) 158
The most common forest site types on mineral soils in Finland are mesic forests (Myrtillus-type), 159
which cover 52% and sub-xeric forests (Vaccinium-type), which cover 26% of forests (Finnish 160
Statistical Yearbook of Forestry 2014).
161
Both forest types in their later successional stages are characterized by dwarf shrubs bilberry 162
(Vaccinium myrtillus L.), lingonberry (Vaccinium vitis-idaea L.) and common heather (Calluna 163
vulgaris L. (Hull)). In sub-xeric forestsV. vitis-idaea andCalluna are dominant, and in mesic forests 164
V. myrtillus is dominant andCalluna practically absent.
165
Managed conifer-dominated mesic and sub-xeric forests on mineral soils typically have an easily 166
distinguishable raw humus layer with a typical thickness of 3-5 cm in Southern Finland (Tamminen 167
1991). In these forests, moss and lichen dominated ground vegetation is the most common and the 168
most important flammable fuel bed, where the majority of forest fires ignite and spread (Schimmel 169
and Granström 1997; Tanskanenet al. 2005). A continuous moss carpet is typical in later 170
successional stages of coniferous forests whereas in young successional stages it is less abundant, thus 171
decreasing fire risk (Schimmel and Granström 1997). Yet, recent clear-cuts where the moss carpet 172
still exists and herbs and graminoids have not yet colonized the areas are flammable similar to the 173
mature forests. A recent study showed that a significant number of forest fires in Sweden are started 174
in clear-cuts as the sparks produced by forest machines are an important source of ignitions (Sjöström 175
et al. 2019). The raw humus layer is also potentially flammable, and the targets and success of 176
prescribed burnings are often estimated by burning depth, which indicates the decrease of moss and 177
raw humus layer.
178
The feather moss (Pleurozium schreberi) (Brid) Mitt. is the most abundant moss species with a 179
coverage of approximately 30% in mesic and 35% in sub-xeric forests. (Mäkipää 2000a). Fork mosses 180
(Dicranumspp., D.polysetum Sw.and D.scopariumHedw. being the most dominant) cover about 10%
181
in both mesic and sub-xeric types (Mäkipää 2000b), whereas stairstep moss (Hylocomium splendens) 182
(Hedw.) is clearly more abundant in mesic types with a share over 10% but in sub-xeric types only 183
3% (Mäkipää 2000c). Reindeer lichens (Cladoniaspp) are practically absent in mesic forests but 184
patchy with an average share of 5% in sub-xeric forests (Nousiainen 2000).Cladonias abundance 185
increases significantly in xeric and barren forests, which are less common (pooled share 4%) and are 186
concentrated in Northern Finland (Finnish Statistical Yearbook of Forestry 2014).
187
188
Twelve forest stands from the study area were chosen, consisting of four different stand types and 189
three replicates from each. The stand types were: 1. Sub-xeric, mature, Pinus dominated stand. 2. Sub- 190
xeric, clear-cut area. 3. Mesic, mature,Picea dominated stand. 4. Mesic, open, clear-cut area (Fig. 1, 191
Table 1). The age and standing stock of a stand is referred to as the developmental stage (either clear- 192
cut or mature) and the combination of forest type and dominant tree species as stand type (either sub- 193
xeric/Pinus or mesic/Picea) (Table 1).
194 195
Table1 196
197
We selected individual stands from the forest planning databases of the study area, according to the 198
following criteria: mature stands had to be over 70 years of age and be either Pinus- orPicea- 199
dominated, with at least 70% dominance (Table 1). The clear-cut stands had to be harvested during 200
the previous winter with no mechanical scarification. All stands had a distinctive raw-humus layer and 201
a characteristic continuous moss layer with patches ofCladonia in sub-xeric stands. The growing 202
stock and structure of the mature stands represented typical Finnish managed forest stands with an 203
evenly aged structure and minor understory.
204 205
From each stand, samples of three dominant moss and/or lichen species were collected on 17 days 206
during summer 2003. The days were chosen using FFI values received from the Finnish 207
Meteorological Institute, so that they would cover different weather and drying conditions (Fig. 2).
208
Sampling was focused especially on dry and drying periods whereas, during constant wet periods 209
(which covered the most part of the sampling period), it was not carried out.
210 211
We sampled each stand in the afternoons of the sampling days. On each occasion, five randomly 212
chosen samples consisting of moss or lichen and raw humus were taken with humus auger with a 213
diameter of 5.8 cm, height of 10 cm and volume of 264 cm3.The samples.were taken from a 300 m2 214
circular sample plot and were located at least 30 m from the stand edge. In mesic stands, the sampled 215
species were:Pleurozium.schreberi, Dicranum spp (D. polysetumbeing the most abundant) and 216
Hylocomium splendens.,and on sub-xeric standsPleurozium,Dicranum andCladonia. (C.
217
rangiferina (L.) Weber ex F.H. Wigg. being the most abundant). The third replication of mesic clear- 218
cut area had an insufficient cover ofHylocomium, so onlyPleuroziumandDicranum were sampled.
219 220
Each sample was then divided into two layers: surface and raw humus. Five subsamples of each layer 221
were pooled into one sample representing the average from that stand. Thus, each sampled stand had 222
six combined samples: a combined sample of each of the three surface species, and three combined 223
samples from raw humus under each species. The collective samples were preserved during 224
transportation in air-tight plastic bags. The fresh-weighing and drying was done directly after 225
transportation with a minimum of 18 hours of oven-drying at 105 ℃. Sufficient drying time was 226
ensured by experimental dryings before actual sampling. After drying, the samples were weighed and 227
the dry-weight FMC was determined.
228 229
Data analysis 230
231
The noon values of FFI and FWI were used in analysis. The FWI values were received from FMI and 232
calculated according to Van Wagner and Pickett (1985) using weather data from the nearest 233
meteorological station located approximately 4 km south-west of the center of the study area. The 234
wind values came from the nearest available station, about 25 km north-east of the study area. We 235
modeled FMC separately for each species, and the surface and raw humus layers, as a function of FFI, 236
stand type, and the development class. Preliminary analyses showed that the shape of the relationship 237
between FMC and FFI varied among the species and was often non-linear. We thus used generalized 238
additive modeling (e.g. Zuur et al. 2009), in which FMC was predicted as a smooth function of FFI.
239
For the strictly positive data (FMC), we used a Gaussian error distribution and log-link function, and 240
the smoothers were allowed to vary as a function of developmental stage. To avoid problems with 241
overfitting and to ensure biologically realistic model behavior, we used monotonically decreasing P- 242
splines as smoothers and limited their flexibility (number of knots in the splinesk = 4). To compare 243
the performance of FFI to the more widely used FWI, we then repeated the analyses, using FWI as the 244
continuous predictor in place of FFI. The models were compared using pseudo-R2 values for both 245
(models with FFI and FWI). For model validation (sensu Zuur et al. 2009), we visually inspected the 246
residuals as a function of FMC and each predictor, as well as day of year to ensure there were no 247
temporal patterns in the residuals (Supplement 2). All models were fitted using R (R Core Team 248
2019) and the package scam (Pya 2018).
249 250
The observed and predicted days of ignition of surface fuels in different stands were analyzed by 251
calculating a probability using FMC frequencies. In Fennoscandia, the FMC values for moisture 252
content of extinction have been estimated to range from 25 to 35 % (Granström and Schimmel 1998;
253
Tanskanenet al. 2005). We used the lower limit since it was considered a more suitable estimate for 254
the timing of prescribed burnings, which was justified because in prescribed burnings one aim is to 255
decrease organic material and ensure a sufficient ecological impact (Lindberget al. 2020). The 256
frequencies over threshold value were compared to all the values of the examined variables or their 257
combinations. Thus, if for instancePleuroziumin sub-xeric clear-cuts had 21 observations under a 258
25% threshold value of FMC, these 21 were compared to all 51 observations in sub-xeric clear-cuts 259
resulting in a probability ratio of 41% (21/51) X 100=41%).
260 261
Results 262
263
During the measurement period, the FMC of surface layer varied between 3% and 300% (Fig. 2). The 264
overall patterns in how the moisture conditions changed during the summer were similar among the 265
species, sites and site types, but the levels differed greatly among species and sites (Fig. 2). It should 266
be noted that the weather conditions during summer 2003 were relatively variable with no long dry 267
periods. This is visible in the distribution of the FFI values, where the highest values (4-6) are 268
missing, which means that the driest circumstances did not occur during sampling (Fig. 2).
269 270
Figure 2 271
272
Of the species,Dicranum was generally the moistest and Cladonia the driest, whereasPleuroziumand 273
Hylocomiumwere between the two. When modeling the FMC as a function of FFI, stand type and 274
developmental stage, several patterns were visible in the surface layer. First, there were clear 275
differences between species in the shape of the relationship between FMC and FFI.Pleurozium, 276
HylocomiumandCladoniahad a tendency for a steadier decline compared toDicranum, which 277
retained moisture up to a higher FFI before declining more rapidly in moisture content (Fig. 3). It is 278
noteworthy that, despite the quick decline at higher FFI values forDicranum, the predicted moisture 279
content in mature stands stayed above the 25-35% level, considered a threshold of ignition (Fig. 3).
280
Stand type was not a significant predictor for any of the species in the surface layer (Table 2). The 281
effect of the developmental stage was significant in the smoother terms only (Table 3, Fig. 4). Plot- 282
level random effects were significant only forPleurozium.
283 284
For the raw humus layer, the relationship between FFI and fuel moisture content were close to linear 285
in most cases, and the differences in the smoothers were clearly smaller compared to the surface layer 286
(Table 2). Similarly, the effect of stand type was different from the surface layer so that, for both 287
PleuroziumandDicranum, the sub-xeric sites were drier than the mesic sites (Table 3). Plot-level 288
random effects were significant only forCladonia. The raw humus variation among the stand types 289
was lower but clear among the developmental stages and, in all stands, well above the 25-35% level.
290 291
Table 2 292
Table 3 293
Figure 3 294
Figure 4 295
296
FWI predicted the FMC of surface layers slightly better than FFI (Table 4). Both models predicted the 297
FMCs ofPleuroziumandHylocomiumbetter than Dicranum andCladonia.In raw humus, the 298
prediction ability was clearly lower, and FWI and FFI performed practically equally (Table 4). The 299
predicted moisture variation curves as a function of FWI are shown in Supplement 3.
300 301
Table 4 302
303
The potential fire hazard days (i.e., days during which the FMC values were under 25%) were highest 304
in Cladoniaand lowest inDicranum(Table 5). Clear-cut areas and sub-xeric pine stands had more 305
fire hazard days than mature stands and mesic spruce-stands. The predicted fire hazard days by FFI 306
formed 6% of sampled days, whereas the observed FMCs of > 25% during the same sampled days 307
was 28%.
308 309
Table 5 310
311
Discussion 312
313
Our results showed that the composition of ground floor vegetation has an effect on the flammability 314
of the surface layer in Fennoscandian boreal forests, and how it varies during the fire season. This 315
flammability was further modulated by the effect of stand growing stock along the lines shown in 316
earlier studies (Granström and Schimmel 1998; Tanskanenet al. 2005; Tanskanenet al. 2006). The 317
differences among species and developmental stages in how the surface layer moisture varied were 318
prominent. As an example,Dicranumin mature stands retained a moisture content well above the 25- 319
35% threshold of the FFI value of 4 (the threshold for public warning), whereasCladoniawas close to 320
the flammability threshold throughout the range of FFI values included in the sample here.
321 322
The development of moisture content between the surface layer and raw humus was clear. Rain 323
usually affects the surface layer saturating it rapidly. The raw humus layer receives some moisture, 324
especially in heavier rains, but dries slowly. However, during longer dry periods, the surface layer and 325
raw humus dry more thoroughly. Long drought periods did not occur during the sampling period so 326
the FMCs in such circumstances could not be compared.
327 328
The FMC variation of surface and raw humus layers was great, especially in higher FMCs, which can 329
be due to several reasons. The same FFI values estimated for a 10 km × 10 km square were used for 330
all stands, so differences in rainfall between stands may have occurred due to local showers. The 331
FMCs were determined layer by layer, which overlooks moisture variation within layers. It is known 332
that the moisture gradient within layers is steep (Vasander and Lindholm 1985), so the upper parts of 333
the surface layer could be clearly drier than the FMCs observed in this study.
334 335
When considering differences among the species in the surface layer,Dicranum was consistently the 336
moistest, andCladonia the driest.Pleurozium andHylocomium were between these two and showed a 337
relatively similar moisture behavior as presented by Busby and Whitefield (1978). The higher FMCs 338
and slower drying curve ofDicranumis probably due to its dense tomentum-covered structure 339
(Peterson and Mayo 1975), which leads to a higher moisture retaining capacity. As reported 340
previously (Mutch and Gastineau 1970; Granström and Schimmel 1998),Cladoniawas the driest 341
surface fuel. This is explained by its gelatinous thallus, loose structure and high surface-to-volume 342
ratio resulting in extreme moisture behavior (Heatwole 1966; Pech 1989, 1991).
343 344
FMC varied among stand types. The results of the FMC variation of the surface layer are in 345
accordance with previous studies in which the differences between stands correlate with their ground 346
vegetation flammability (Tanskanenet al.2006). Using 30% threshold values for the FMC of moss 347
layer, Tanskanenet al. (2006) reported two times more potential days of ignition in open than in 348
mature areas, and inPinus-dominated stands two to three times higher than in Picea-dominated 349
stands. In our study, the differences between clear-cut and mature developmental stages were clear, 350
but the impact of site type and the associated dominant tree species was smaller.
351
Comparison between the Finnish FFI and Canadian FWI showed that FWI was consistently a better 352
predictor for the moisture content of the surface layer fuels, irrespective of the species. For the raw 353
humus layer, the two indices performed almost identically. The better performance of FWI for surface 354
fuels was similar to what Tanskanenet al. (2005) reported. Thus the CFFWIS could well be used in 355
Finland.
356
Our results support the conclusions of Tanskanenet al. (2005) and Vajdaet al. (2014) suggesting that 357
FFI could be improved by using forest stand variables. Such parameters as developmental stage and 358
dominant tree species could likely improve the FFIs prediction ability significantly, which could 359
eventually help practical fire suppression activities by better anticipation and preparation.
360
Fire history studies in Fennoscandia have reported great variation in fire cycles. The shorter cycles 361
have been typical inPinus-dominated forests, especially in south- and middle boreal forests (e.g., 362
Lehtonen and Kolström 2000), whereas in more northern and Picea-dominated forests, the cycle has 363
been longer (e.g. Wallenius 2004). The differences have been explained by meteorological factors, 364
dominant tree species, vegetation, fire suppression and general human influence (Wallenius 2004, 365
2011). According to our results, the differences in reported fire cycles could be partially explained by 366
dominant tree species and changes in ground floor vegetation, especially in lichen-bryophyte ratio.
367
For example, the abundance ofCladonia has substantially decreased in recent decades in Finland 368
(Nousiainen 2000; Mäkipää and Heikkinen 2003; Tonteriet al. 2013). At the same time, a notable 369
increase in the abundance ofDicranumhas been documented especially in Northern Finland 370
(Mäkipää 2000b). It is possible that reduction in the cover of fast-dryingCladoniaand increase in the 371
cover of slowly-dryingDicranumhas partially reduced forest fire risk particularly in Northern 372
Finland.
373 374
In our study, the large variation of FMC in different stands and ground floor fuel materials show that 375
potential days for prescribed burnings also have a large variation, especially when the variable 376
ecological targets of burnings are taken into account. An often presented rule of thumb in guidelines 377
for prescribed burnings is that the forest fire warning in Finland (FFI value 4) could be considered as 378
a general threshold for successful burnings (Lemberg and Puttonen 2002). According to our results, 379
this assumption is too simplistic, since suitable days for prescribed burning also seem to occur with 380
lower FFI values. Yet it should be noted that the selected level of FMC 25% should be interpreted as a 381
level where burning of studied surface layer fuels is possible. Thus, the various goals of prescribed 382
burnings should be taken into account when suitable burning conditions are determined. For instance, 383
in most restoration burnings no special burning depth is targeted as it is in silvicultural burnings. On 384
the other hand, denser stands where restoration burnings are performed dry slower than regeneration 385
areas. Also, if the aim is also to burn the humus layer, long drought periods are needed since the FMC 386
values of raw humus did not reach the ignition threshold limits within the range of the FFI values we 387
analyzed. Thus, a stand-specific monitoring of surface fuel and raw humus layer is recommended so 388
that all potential burning days – whose small number often functions as a limiting factor – could be 389
utilized more effectively, and the targeted impacts of burnings could be ensured.
390 391
Conclusions 392
Our results show that the different ground vegetation fuels differ in their moisture variation and 393
ignition potential. Developmental stage and stand type of the forest affect the moisture variation of the 394
studied fuels. Canadian FWI predicted the FMC of surface layer better than Finnish FFI, so it could be 395
used in Finland. We conclude that, by using additional predictor variables, the ability of forest fire 396
indices to predict fuel moisture could be improved. This could benefit forest fire prevention by 397
enhancing early warning systems and by developing a GIS-based system providing online stand-wise 398
FMC estimates of surface fuels, which could be utilized in practical firefighting as well as in 399
prescribed burning.
400 401
Abbreviations 402
CFFWIS Canadian Forest Fire Weather Index System 403
DC Drought Code 404
DMC Duff Moisture Code 405
FFI Finnish Forest Fire Index 406
FFMC Fine Fuel Moisture Code 407
FMC Fuel moisture content 408
FMI Finnish Meteorological Institute 409
FWI Canadian Fire Weather Index 410
411 412
Acknowledgements 413
We thank Antti Kujala and Tuija Toivanen for outstanding field work, Ilari Lehtonen and Ari 414
Venäläinen for providing us FFI and FWI data as well as valuable information on FFI and Sanna 415
Laaka-Lindberg for commenting the manuscript.
416 417
Conflicts of interest: The authors declare no conflicts of interest.
418 419
Declarations of Funding:The data collection of study was partially funded by European Union Fifth 420
Framework projects SPREAD and EUFIRELAB. TA was funded by the Kone Foundation.
421 422 423 424 425 426
References 427
428
Ahti T, Hamet-Ahti L, Jalas J (1968) Vegetation zones and their sections in 429
northwestern Europe.Annales Botanici Fennici 5, 169–211.
430
Busby JR, Whitfield, DWA (1978) Water potential, water content, and net assimilation of some 431
boreal forest mosses.Canadian Journal of Botany 56, 1551–58. https://doi.org/10.1139/b78- 432
433 184
Dimitrakopoulos AP, Bemmerzouk AM, Mitsopoulos ID (2011) Evaluation of the Canadian fire 434
weather index system in an eastern Mediterranean environment. Meteorological. Applications 435
18, 83–93. https://doi.org/10.1002/met.214 436
Ferguson SA, Ruthford JE, McKay SJ, Wright D, Wright C, Ottmar R (2002) Measuring moisture 437
dynamics to predict fire severity in longleaf pine forests. International Journal of Wildland 438
Fire 11(4), 267–279. https://doi.org/10.1071/WF02010 439
Finnish Forestry - Practice and Management (2011) Metsäkustannus, Helsinki. 271 p.
440
Finnish statistical yearbook of forestry 2014 (2014) Finnish Forest Research Institute.
441
http://urn.fi/URN:ISBN:978-951-40-2506-8 442
Granström A, Schimmel J (1998) Utvärdering av det kanadensiska brandrisksystemet – 443
testbränningar och uttorkningsanalyser. (In Swedish with English abstract: Assessment of the 444
Canadian forest fire danger rating system for Swedish fuel conditions.) P21–244/98. (Rescue 445
Service: Karlstad, Sweden) 446
Heatwole H (1966) Moisture exchange between the atmosphere and some lichens of the genus 447
Cladonia.Mycologia 58, 148-156. Available at https://www.jstor.org/stable/3756996 448
[Verified 28 November 2020].
449
Heikinheimo M, Venäläinen A, Tourula T (1998) A soil moisture index for the assessment of 450
forest fire potential in the boreal zone. In Proceedings of the International Symposium on 451
Applied Agrometeorology and Agroclimatology (Volos, Greece), Office for Official 452
Publication of the European Commission (Luxembourg), NR Dalezios (ed), EUR 18328‐
453
COST 77, 79, 711. 549– 555.
454
Hille MG, den Ouden J (2005) Fuel load, humus consumption and humus moisture dynamics in 455
Central European Scots pine stands.International Journal of Wildland Fire 14, 153-159 456
https://doi.org/10.1071/WF04026 457
Hille MG, Stephens S (2005) Mixed Conifer Forest Duff Consumption during Prescribed Fires:
458
Tree Crown Impacts.Forest Science 51(5), 417-424. Available at 459
https://nature.berkeley.edu/stephenslab/wp-content/uploads/2015/04/Hille-Stephens-duff-FS- 460
[Verified 28 November 2020].
461
Juvakka M, Viinikainen J, Puputti I, Kuupakko S (1995) Vesijaon tutkimusalue, hoito- ja 462
käyttösuunnitelma 1994–2003. [Plan for the management and use of forests in Vesijako 463
research area 1994–2003]. Metlan tutkimusmetsien julkaisusarja 5. Vantaa. 228 p. ISSN 464
1238-0830. (In Finnish).
465
Kilpeläinen A, Kellomäki S, Strandman H, Venäläinen A (2010) Climate change impacts on 466
forest fire potential in boreal conditions in Finland.Climatic Change 103, 383–398 467
https://doi.org/10.1007/s10584-009-9788-7 468
Kuuluvainen T, Lindberg H, Vanha-Majamaa I, Keto-Tokoi P, Punttila P (2019) Low-level 469
retention forestry, certification and biodiversity: case Finland.Ecological Processes 8,47.
470
https://doi.org/10.1186/s13717-019-0198-0 471
Lehtonen H, Kolström T (2000) Forest fire history in Viena Karelia, Russia.Scandinavian 472
Journal of Forest Research 15, 585-590. https://doi.org/10.1080/02827580050216833 473
Lehtonen I, Ruosteenoja K, Venäläinen A, Gregow H (2014) The projected 21st century forest 474
fire risk in Finland under different greenhouse gas scenarios.Boreal Environment Research 475
19, 127-139. Available at:
476
https://www.researchgate.net/publication/285955800_The_projected_21st_century_forest- 477
fire_risk_in_Finland_under_different_greenhouse_gas_scenarios [Verified 28 November 478
2020].
479
Lehtonen I, Venäläinen A, Kämäräinen M, Peltola H, Gregow H (2016) Risk of large-scale fires 480
in boreal forests of Finland under changing climate. Natural Hazards and Earth System 481
Sciences 16, 239-–253. https://doi.org/10.5194/nhess-16-239-2016 482
Lemberg T, Puttonen P (2002) Kulottajan käsikirja. Metsälehti kustannus. Vammalan kirjapaino.
483
(Guide for prescribed burning. Textbook. In Finnish) 484
Lindberg H, Punttila P, Vanha-Majamaa I (2020) The challenge of combining variable retention 485
and prescribed burning in Finland.Ecological Processes 9, 4 (2020).
486
https://doi.org/10.1186/s13717-019-0207-3 487
Mäkelä HM, Laapas M, Venäläinen, A (2012) Long-term temporal changes in the occurrence of a 488
high forest fire danger in Finland.Natural Hazards and Earth System Sciences 12, 2591-2601 489
https://doi.org/10.5194/nhess-12-2591-2012 490
Mäkelä HM, Venäläinen A, Jylhä K, Lehtonen I, Gregow H (2014) Probabilistic 491
projections of climatological forest fire danger in Finland.Climate Research 60, 73-85.
492
Available at https://www.jstor.org/stable/24896175 [Verified 28 November 2020].
493
Mäkipää R (2000a)Pleurozium schreberi. In: Reinikainen A, Mäkipää R, Vanha-Majamaa I, 494
Hotanen J-P. (eds.) 2000. Kasvit muuttuvassa metsäluonnossa. [Summary in English:
495
Changes in the frequency and abundance of forest and mire plants in Finland since 1950].
496
Tammi, Jyväskylä. 384 p.
497
Mäkipää R (2000b)Dicranum. In: Reinikainen A, Mäkipää R, Vanha-Majamaa I, Hotanen J-P.
498
(eds.) 2000. Kasvit muuttuvassa metsäluonnossa. [Summary in English: Changes in the 499
frequency and abundance of forest and mire plants in Finland since 1950]. Tammi, Jyväskylä.
500
384 p.
501
Mäkipää R (2000c)Hylocomium splendensi. In: Reinikainen A, Mäkipää R, Vanha-Majamaa I, 502
Hotanen J-P. (eds.) 2000. Kasvit muuttuvassa metsäluonnossa. [Summary in English:
503
Changes in the frequency and abundance of forest and mire plants in Finland since 1950].
504
Tammi, Jyväskylä. 384 p.
505
Mäkipää R, Heikkinen J (2003) Large-scale changes in abundance of terricolous bryophytes 506
and macrolichens in Finland.Journal of Vegetation Science 14, 497–508.
507
https://doi.org/10.1111/j.1654-1103.2003.tb02176.x 508
Mutch, RW, Gastineau OW (1970) Timelag and equilibrium moisture content of reindeer lichen.
509
Res. Pap. INT-76. Ogden, UT: U.S. Department of Agriculture, Forest Service, Intermountain 510
Forest and Range Experiment Station. 8 p. https://doi.org/10.5962/bhl.title.68840 511
Nousiainen H (2000)Cladina In: Reinikainen A, Mäkipää R, Vanha-Majamaa I, Hotanen, J-P.
512
(eds.) 2000. Kasvit muuttuvassa metsäluonnossa. [Summary in English: Changes in the 513
frequency and abundance of forest and mire plants in Finland since 1950]. Tammi, Jyväskylä.
514
384 p.
515
Okko M (1972) Jäätikön häviämistapa Toisen Salpausselän vyöhykkeessä Lammilla.
516
Summary: Deglaciation in the Second Salpausselka ice-marginal belt at Lammi, South 517
Finland.Terra 84(3), 115-123.
518
Päätalo ML (1998) Factors influencing occurrence and impacts of fires in Northern European 519
forests.Silva Fennica 32(2), 185-202. https://doi.org/10.14214/sf.695 520
Péch G (1989) A model to predict the moisture content of reindeer lichen.Forest Science 35, 521
1014-1028. https://doi.org/10.1093/forestscience/35.4.1137 522
Péch G (1991) Dew on reindeer lichen.Canadian Journal of Forest Research 21, 1415–1418.
523
https://doi.org/10.1139/x91-198 524
Peterson W, Mayo J (1975) Moisture stress and its effect on photosynthesis inDicranum 525
polysetum.Canadian Journal of Botany 53, 2897–2900. Available at 526
https://fdocuments.in/document/moisture-stress-and-its-effect-on-photosynthesis-in- 527
dicranum-polysetum.html [Verified 28 November 2020].
528
Pya N (2018) scam: Shape Constrained Additive Models. R package version 1.2-3.
529
R Core Team (2019) R: A language and environment for statistical computing. Version 3.5. R 530
Foundation for Statistical Computing, Vienna, Austria.
531
Sandberg DV (1980) Duff reduction by prescribed burning in Douglas-fir. USDA Forest Service 532
Research Paper PNW-272, 18 p. https://doi.org/10.2737/PNW-RP-272 533
San-Miguel-Ayanz J, Carlson JD, Alexander M, Tolhurst K, Morgan G, Sneeuwjagt R, Dudley M 534
(2003) Current methods to assess fire danger potential. In Wildland Fire Danger Estimation 535
and Mapping.The Role of Remote Sensing Data, Chuvieco E (ed). World Scientific 536
Publishing: Singapore; 21–61. Available at 537
https://pdfs.semanticscholar.org/da16/f999aff0083cfee5820cfad43a8d6d1e4c41.pdf [Verified 538
28 November 2020].
539
Schimmel J, Granström A (1997) Fuel succession and fire behavior in the Swedish boreal forest.
540
Canadian Journal of Forest Research 27, 1207–1216. https://doi.org/10.1139/x97-072 541
Sjöström J, Plathner FV, Granström A (2019) Wildfire ignition from forestry machines in boreal 542
Sweden.International Journal of Wildland Fire 28(9), 666-677.
543
https://doi.org/10.1071/WF18229 544
Tamminen P (1991) Kangasmaan ravinnetunnusten ilmaiseminen ja viljavuuden alueellinen 545
vaihtelu. Summary: Expression of soil nutrient status and regional variation in soil fertility of 546
forested sites in southern Finland.Folia Forestalia 777. 40 p. http://urn.fi/URN:ISBN:951- 547
40-1170-8 548
Tanskanen H, Venäläinen A, Puttonen P, Granström A (2005) Impact of stand structure on 549
surface fire ignition potential inPicea abies andPinus sylvestris forests in southern Finland.
550
Canadian Journal of Forest Research 35, 410-420. https://doi.org/10.1139/X04-188 551
Tanskanen H, Granström A, Venäläinen A, Puttonen P (2006) Moisture dynamics of moss 552
dominated surface fuel in relation to the structure ofPicea abies andPinus sylvestris stands.
553
Forest Ecology and Management.226, 189-198. doi:10.1016/j.foreco.2006.01.048 554
Tonteri T, Salemaa M, Rautio P (2013) Changes of understorey vegetation in Finland in 1985–
555
2006. In: Merilä, P. & Jortikka, S. (eds.). Forest Condition Monitoring in Finland – National 556
report. The Finnish Forest Research Institute. [Online report].
557
http://urn.fi/URN:NBN:fi:metla-201305087583.
558
Vajda A, Venäläinen A, Suomi I, Junila P, Mäkelä HM (2014) Assessment of forest fire danger in 559
a boreal forest environment: description and evaluation of the operational system applied in 560
Finland.Meteorological Applications 21, 879-887. https://doi.org/10.1002/met.1425 561
Van Wagner CE (1987) Development and Structure of the Canadian Forest Fire Weather Index 562
System; Forestry Technical Report 35; Canadian Forestry Service: Ottawa, ON, Canada, 563
Volume 1, 48 p. Available at https://cfs.nrcan.gc.ca/publications?id=19927 [Verified 28 564
November 2020].
565
Van Wagner CE, Pickett TL (1985) Equations and FORTRAN program for the Canadian Forest 566
Fire Weather Index System. Canadian Forestry Service, Petawawa National Forestry Institute, 567
Chalk River, Ontario. Forestry Technical Report 33. 18 p. Available at 568
https://cfs.nrcan.gc.ca/publications?id=19973 [Verified 28 November 2020].
569
Vasander H, Lindholm T (1985) Fire intensities and surface temperatures during prescribed 570
burning.Silva Fennica 19(1), 1-15. https://doi.org/10.14214/sf.a15406 571
Venäläinen A, Heikinheimo M (2003) The Finnish forest fire index calculation system. In:
572
Zschau, J. & Kuppers, A. (ed.). Early Warning Systems for Natural Disaster Reduction.
573
Springer, 467 p.
574
Wallenius T (2004) Fire histories and tree ages in unmanaged boreal forests in Eastern 575
Fennoscandia and Onega peninsula. Academic Dissertation, June 2004.University of Helsinki, 576
Faculty of Biosciences, Department of Biological and Environmental Sciences and Faculty of 577
Agriculture and Forestry, Department of Forest Ecology. http://urn.fi/URN:ISBN:952-10- 578
1893-3 579
Wallenius T (2011) Major decline in fires in coniferous forests - reconstructing the phenomenon 580
and seeking for the cause.Silva Fennica 45, 139-155. https://doi.org/10.14214/sf.36 581
Yearbook of forest statistics 1990-1991 (1992) Finnish Forest Resource Institute.
582
http://urn.fi/URN:ISBN:951-40-1205-4 583
Ziel RH, Bieniek PA, Bhatt US, Strader H, Rupp, TS, York AA (2020) Comparison of Fire 584
Weather Indices with MODIS Fire Days for the Natural Regions of Alaska.Forests 11(5), 585
516; https://doi.org/10.3390/f11050516 586
Zuur A, Ieno EN, Walker N, Saveliev AA, Smith, GM (2009) Mixed Effects Models and 587
Extensions in Ecology with R. New York, NY. Springer.
588 589 590 591
Tables and figure captions 592
593
Table 1. The sampled stands. In clear-cut areas the dominant tree species refers to species of the pre- 594
cut stand. Pine: Pinus sylvestris, spruce:Picea abies,birch: Betulaspp.
595 596
Stand Developmental stage Stand type Age,
years
Average height, meters
Standing stem volume: cubic meters/hectare
Standing tree species percentages by volume (pine/spruce/birch)
SXC1 clear-cut sub-xeric/pine 0 0 0 -
SXC2 clear-cut sub-xeric/pine 0 0 0 -
SXC3 clear-cut sub-xeric/pine 0 0 0 -
SXM1 mature sub-xeric/pine 90 24 210 90/10/0
SXM2 mature sub-xeric/pine 120 26 250 100
SXM3 mature sub-xeric/pine 120 25 240 100
MC1 clear-cut mesic/spruce 0 0 0 -
MC2 clear-cut mesic/spruce 0 0 0 -
MC3 clear-cut mesic/spruce 0 0 0 -
MM1 mature mesic/spruce 75 26 260 10/80/10
MM2 mature mesic/spruce 90 28 310 10/90/0
MM3 mature mesic/spruce 90 27 290 10/90/10
597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
Table 2. Parametric coefficients for factor variables in the models. Estimates for the developmental 623
stage (Dev. Stage) are relative to clear-cut area, and site type relative to mesic site type.Hylocomium 624
andCladoniaoccurred only on a single type.
625 626
Layer Species Variable Estimate Std. Error t p
Surface Pleurozium Intercept 4.75 1.72 2.76 0.006 **
Dev. stage mature forest 2.24 2.42 0.92 0.356
Site type sub-xeric -0.23 0.16 -1.47 0.144
Surface Dicranum Intercept 5.00 0.90 5.56 < 0.001 ***
Dev. stage mature forest 0.58 0.91 0.64 0.523
Site type sub-xeric -0.10 0.10 -0.98 0.327
Surface Hylocomium Intercept 3.98 0.23 17.36 < 0.001 ***
Dev. stage mature forest 2.85 2.54 1.12 0.266
Surface Cladonia Intercept 3.63 0.17 21.11 < 0.001 ***
Dev. stage mature forest 2.13 2.58 0.82 0.412
Raw humus Pleurozium Intercept 5.39 0.21 26.01 < 0.001 ***
Dev. stage mature forest 0.13 0.35 0.36 0.719
Site type sub-xeric -0.20 0.05 -4.24 < 0.001 ***
Raw humus Dicranum Intercept 4.96 0.41 12.16 < 0.001 ***
Dev. stage mature forest 0.73 0.51 1.43 0.156
Site type sub-xeric -0.16 0.06 -2.44 0.016 *
Raw humus Hylocomium Intercept 5.30 0.37 14.50 < 0.001 ***
Dev. stage mature forest 0.43 0.47 0.92 0.363
Raw humus Cladonia Intercept 4.76 0.09 54.53 < 0.001 ***
Dev. stage mature forest 0.62 0.28 2.24 0.027 *
Significant variables (p < 0.05) are in bold
627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642
Table 3. Significance of smoother terms and plot-level random effects 643
644 645
Layer Species Smoother term F p
Surface Pleurozium s(FFI) x Dev. stage clearcut 32.18 < 0.001 ***
s(FFI) x Dev. stage mature forest 27.33 < 0.001 ***
plot (random effect) 3.09 < 0.001 ***
Dicranum s(FFI) x Dev. stage clearcut 27.37 < 0.001 ***
s(FFI) x Dev. stage mature forest 29.96 < 0.001 ***
plot (random effect) 0.04 0.393
Hylocomium s(FFI) x Dev. stage clearcut 15.18 < 0.001 ***
s(FFI) x Dev. stage mature forest 12.75 < 0.001 ***
plot (random effect) 0.31 0.326
Cladonia s(FFI) x Dev. stage clearcut 28.54 < 0.001 ***
s(FFI) x Dev. stage mature forest 11.76 < 0.001 ***
plot (random effect) 0.00 0.841
Raw humus Pleurozium s(FFI) x Dev. stage clearcut 11.07 < 0.001 **
s(FFI) x Dev. stage mature forest 2.49 0.111
plot (random effect) 0.19 0.366
Dicranum s(FFI) x Dev. stage clearcut 5.93 0.004 **
s(FFI) x Dev. stage mature forest 2.36 0.118
plot (random effect) 1.73 0.023 *
Hylocomium s(FFI) x Dev. stage clearcut 3.66 0.060
s(FFI) x Dev. stage mature forest 6.77 0.011 *
plot (random effect) 0.00 0.815
Cladonia s(FFI) x Dev. stage clearcut 30.09 < 0.001 ***
s(FFI) x Dev. stage mature forest 5.18 0.026 *
plot (random effect) 2.84 0.017 *
Significant variables (p < 0.05) are in bold
646 647 648 649 650 651
Table 4. Performance of the Finnish Forest Fire Index (FFI) compared to the Canadian Fire Weather 652
Index (FWI) as a predictor of FMC in different layers, measured as pseudo-R2.
653
654
Surface layer FFI FWI
R2 R2
Pleurozium 0.55 0.64
Dicranum 0.46 0.54
Hylocomium 0.6 0.69
Cladonia 0.45 0.52
Raw humus FFI FWI
R2 R2
Pleurozium 0.26 0.25
Dicranum 0.36 0.34
Hylocomium 0.35 0.36
Cladonia 0.42 0.34
655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
Table 5. The potential fire hazard days (defined as fuel moisture content values under 25%) of studied 671
surface layer materials, stand types and developmental stages. (MT= mesic stand, SX= sub-xeric 672
stand, C=clear-cut area, M=mature stand, FFI pred = the potential days of ignition predicted by 673
Finnish Forest Fire Index (FFI), index values> 4) 674
675 676
MTC MTM SXC SXM MT SX C M FFFI pred Total
Pleurozium 54 % 8 % 41 % 31 % 28 % 36 % 47 % 20 % 6 % 32 %
Dicranum 32 % 0 % 27 % 8 % 14 % 18 % 29 % 4 % 6 % 16 %
Hylocomium 54 % 4 % 22 % 54 % 4 % 6 % 22 %
Cladonia 71 % 20 % 45 % 71 % 20 % 6 % 45 %
Total 45 % 4 % 46 % 20 % 21 % 33 % 46 % 12 % 6 % 28 %
FFI > 4 6 %
FFI < 4 94 %
677 678 679 680
Figure 1. Location of sampled stands 681
Figure 2. The observed fuel moisture contents (FMC) and Finnish Forest Fire Index (FFI) values on 682
sampling days. Note the different y-axes.
683
Figure 3. The predicted fuel moisture content (%) of each studied species, by stand type and 684
developmental stage, as a function of Finnish Forest Fire Index (FFI). Dotted lines show the 25-35%
685
moisture content.
686
Figure 4. The predicted fuel moisture content (%) by studied species, as a function of Finnish Forest 687
Fire Index (FFI) on different stand types and developmental stages. Dotted lines show the 25-35%
688
moisture content.
689 690
691
Figure 1.
692 693 694 695
696 697
Figure 2.
698 699 700 701
702 703
Figure 3.
704
705 706
Figure 4.
707 708